MiniMax releases M3 with MSA architecture, 1M-token context, native multimodality, and frontier-level coding and agentic performance.

MiniMax Releases MMX-CLI: A Command-Line Interface That Gives AI Agents Native Access to Image, Video, Speech, Music, Vision, and Search

The Releasebot-curated entry on the July 31, 2026 MiniMax H3 launch supplies substantive technical detail beyond the release-notes index. H3 is described as a general-purpose omni-modal generation model that jointly understands text, images, video, and audio, and generates video with native stereo audio at up to 2K res The page also reports that MiniMax plans to open up the H3 model weights in the coming days subject to applicable laws and regulations, a hedging clause that leaves release timing and jurisdiction-dependent availability unconfirmed. Hardware compatibility was a stated design priority from early in H3's development, aim

The first-party MiniMax platform release notes index enumerates a chronological model timeline through July 31, 2026, providing the canonical roadmap for the provider. The most recent entry is MiniMax H3 (Jul 31, 2026), described as a new-generation open general-purpose multimodal video model that handles text, image, This page is a first-party documentation index rather than a full changelog, so per-model API endpoints, rate limits, and deprecations are not captured in the supplied excerpt. Capability and SOTA claims are vendor self-reported and unverified by independent benchmarks. The URL host is platform.minimax.io while the pro